Observations
First, although I have no abilities, I have no trouble believing that a skilled shooter could influence the outcome of a throw of the dice. Why not? Humans are able to train themselves to achieve high levels of proficiency at all kinds of physical activities. And I'm not speaking only of elite athletes. With dedication and training, plenty of weekend warriors advance quickly in sports such as skiing, tennis, golf, and pool. The skill of throwing dice in a standardized motion should be no different.
But the key word here is "influence." No one (at least no one reputable) claims to be able to control a throw to the point of hitting a certain number every time, or even to the point of being able to prevent the dice from going off-axis any overwhelming percentage of the time. Instead, the goal is keep the dice on axis with a frequency sufficient (compared to a random roll) to modestly alter the distribution of numbers thrown.
The question becomes, is the achievable level of influence sufficient to make craps an attractive advantage play? Well, if one is simply making pass line bets, taking no odds, one has to overcome a basic house edge of about 1.41%. It's worth noting that this house edge is a lot higher than for blackjack games that knowledgeable players will generally attack through card counting. If the rules were so bad that a blackjack game had a basic strategy house edge of 1.41%, counters would almost certainly spurn it and look for something better. So one needs a pretty high level of influence to overcome the basic house edge on the pass line bet.
On the other hand, craps has the unique feature of allowing players to make odds bets, that pay true odds, with no house advantage. Many casinos will allow an odds bets as high as 10 or 20 times the flat bet, sometimes even more. Clearly, it requires much less influence over the outcome of a throw in order to achieve an advantage at odds bets. If the odds bets one is making are very large compared to the flat bets, the results of the flat bets become relatively insignificant compared to the results of the odds bets. With only modest influence over the results of throws of the dice, it soon becomes possible to achieve an overall edge in the game. This key role of odds bets is a point that has escaped some who have written in recent threads.
But there is a serious down side to the strategy of making large odds bets. They hugely increase the variance of the play. The variance of a hand of blackjack is something like 1.2 times one's initial bet. The variance of a pass line bet backed up with maximum odds can be astronomically higher. Whether one looks to the Kelly betting criterion, or to a risk of ruin analysis, one finds that bankroll requirements are going to be very high for any given level of betting. And the average pass line bet takes much longer to resolve than the average hand of blackjack. Whereas playing blackjack, a rate of 100 hands per hour would be nothing special, you might be lucky to get 20 pass line bets per hour playing craps. This implies a depressingly low expected hourly win rate, for any given bankroll. And finally, the combination of a low number of bets per hour and high variance will mean that N0--one measure of the time required to get into the long run--will be huge.
So I have little doubt that one can achieve an edge in craps. But I'm highly dubious that the level of the edge that's achievable is good enough, taking variance into account, to make dice control a reliable way to make a steady income, or to make it especially attractive compare to other advantage plays that are readily available.